TranceAddict Forums

TranceAddict Forums (www.tranceaddict.com/forums)
- Production Studio
-- Comparing CD vs mp3 quality: which frequencies are lost in mp3s?!..
Pages (2): [1] 2 »


Posted by Dance123 on Dec-26-2004 18:21:

Question Comparing CD vs mp3 quality: which frequencies are lost in mp3s?!..

Hi,

When comparing quality of commercial CDs vs mp3s (even at 192kbps), I notice that the mp3s sound much more dull then the original CD. I also have the impression that the bass frequencies are much louder, too heavy, muddy with mp3s compared to the original CD. Anybody else notices this or is it just my imagination? I listen music mainly on decent hifi speakers.

Since mp3 is a compression format, isn't it so that certain frequencies are being cut out or something?! How does that work with mp3's, like which frequencies exactly are being cut out?! Anybody who knows more about this could please explain this? Also, how come that the bass sounds heavier, or is this just my imagination?

Thanks!


Posted by robin on Dec-26-2004 19:08:

http://www.iis.fraunhofer.de/ *hope I got the url right *

I bet the info is on that page somewere, I wouldn't know myself tbh.


Posted by Doogee on Dec-26-2004 20:26:

http://en.wikipedia.org/wiki/Mp3

That has everything you need to know...


Posted by Derivative on Dec-27-2004 02:19:

it cuts out alot of audible and semi inaudible frequencies. the extent to which is does this is determined by the bitrate its encoded at.

i think even with 320 kbps mp3 the freq range reproduction is between 15hz and about 15,500hz or something like that. thats why the most noticeable differences between mp3 and CDDA are often in the hihats. or at least, most people notice flatter, tinnier hihats with mp3s because the frequency response doesnt extend that far. it loses some sub bass response too but it depends on the record and the amount of bass presence. at high levels certain bass heavy recordings just lose something noticeably when played on PAs at club levels - even at 320 kbps. even though you cant physically hear ultra high and low frequency sound you can sometimes notice when its not there. case in point - have someone blow a dog whistle near you.

then theres the fact that your speakers/headphones have an effective frequency response too. most people dont have audio equipment sensitive enough to pick up frequencies above 15,000 hz so they wouldnt notice any degradation in sound quality on say most mini hifi systems. i have some sony hifi speakers which are pretty bass heavy and the freq response drops off pretty heavily and sharply around teh upper mids. i reckon 320 kpbs mp3s would be largely indistinguishable from CDDA on these speakers simply because they arent sensitive enough. on monitors you notice the difference more. on really sensitive monitoring speakers and PAs and high levels it is sometimes clearly audible.


Posted by Dance123 on Dec-27-2004 04:24:

If the freq range of mp3s (even at 320kbps) starts at 15Hz, doesn't give you that enough subbass frequencies? I actually got the impression that with mp3s the bass is too heavy (or more muddy) compared to the original CD.

Also, is it really so that mp3s only go to 15khz?! It would explain why mp3s don't sound so bright as CDs when listening on some decent hifi speakers, however isn't it so that mp3s are also very weak in the mid frequencies. Isn't it so that mp3's cut frequencies in the entire frequency range in order to make the file smaller (compression)or how does that work?


Posted by State of Matter on Dec-27-2004 04:44:

I'm sure it cuts a little from the entire spectrum but it concentrates on the high frequencies for sure. Listen to a 96 kbps mp3, bass will generally sound fine, albeit a bit muddy, but anything over like 5 khz is completely shot. Anyways, a lot of dj's play 192 to 320 kbps mp3's, and they sound fine at the clubs. The only way you hear the difference is if you put them right next to eachother, A/B, but for all practical purposes a well mastered hi fi mp3 will suffice behind the decks just as well as any wav.


Posted by ZxZDeViLZxZ on Dec-27-2004 18:06:

you know comparing cd to vinyl tho you loose even more sound... honestly the best sound youll ever get is from vinyl. not even dvd sound can compare to vinyl look at chart below:





so really if your going to take a quality loss why not just use mp3 becuase they all loose quality when compared to a vinyl heres the full report below..... taken from how stuff works





Is the sound on vinyl records better than on CDs or DVDs?

The answer lies in the difference between analog and digital recordings. A vinyl record is an analog recording, and CDs and DVDs are digital recordings. Take a look at the graph below. Original sound is analog by definition. A digital recording takes snapshots of the analog signal at a certain rate (for CDs it is 44,100 times per second) and measures each snapshot with a certain accuracy (for CDs it is 16-bit, which means the value must be one of 65,536 possible values).

Comparison of a raw analog audio signal to the CD audio and DVD audio output



This means that, by definition, a digital recording is not capturing the complete sound wave. It is approximating it with a series of steps. Some sounds that have very quick transitions, such as a drum beat or a trumpet's tone, will be distorted because they change too quickly for the sample rate.

In your home stereo the CD or DVD player takes this digital recording and converts it to an analog signal, which is fed to your amplifier. The amplifier then raises the voltage of the signal to a level powerful enough to drive your speaker.

A vinyl record has a groove carved into it that mirrors the original sound's waveform. This means that no information is lost. The output of a record player is analog. It can be fed directly to your amplifier with no conversion.

This means that the waveforms from a vinyl recording can be much more accurate, and that can be heard in the richness of the sound. But there is a downside, any specks of dust or damage to the disc can be heard as noise or static. During quiet spots in songs this noise may be heard over the music. Digital recordings don't degrade over time, and if the digital recording contains silence, then there will be no noise.

From the graph above you can see that CD quality audio does not do a very good job of replicating the original signal. The main ways to improve the quality of a digital recording are to increase the sampling rate and to increase the accuracy of the sampling.

The recording industry has a new standard for DVD audio discs that will greatly improve the sound quality. The table below lists the sampling rate and the accuracy for CD recordings, and the maximum sampling rate and accuracy for DVD recordings. DVDs can hold 74 minutes of music at their highest quality level. CDs can also hold 74 minutes of music. By lowering either the sampling rate or the accuracy, DVDs can hold more music. For instance a DVD can hold almost 7 hours of CD quality audio.

CD Audio DVD Audio
Sampling Rate 44.1 kHz 192 kHz
Samples per second 44,100 192,000
Sampling Accuracy 16-bit 24-bit
Number of Possible Output Levels 65,536 16,777,216


DVD audio discs and players are rare right now, but they will become more common, and the difference in sound quality should be noticeable. To take advantage of higher quality DVD audio discs, however, you will need a DVD player with a 192kHz/24-bit digital to analog converter. Most DVD players only have a 96kHz/24-bit digital to analog converter. So if you are planning to take full advantage of DVD audio be sure to look for a 192kHz/24-bit DAC.


Posted by Pjotr G on Dec-27-2004 18:32:

wow, I really hate it when websites that people use as a resource are full of crock.

Where does the actual waveform being cut onto the record come from huh? Usually DAT or CD. Both digital formats. Plus you get the fact that mastering engineers have to cut high end before the record can be cut to avoid problems such as the needle being too big to handle a super narrow groove. In practice, this cut is lower than the specs of CD. Oh, and the actual waveform is cut on the record, uhh whatever.

Sprry vinyl freaks. I love vinyl too. But it's not because of scientific bullshit. It has a specific nice sound to it. But it's by no means a more accurate representation of an original master, than, say, CD.


Posted by ZxZDeViLZxZ on Dec-27-2004 22:39:

show me stats and where you find them then lets see this.... i hate when people post replys and cant even included any facts kinda like trying to smoking a joint without having weed


Posted by Pjotr G on Dec-27-2004 23:24:

ok here's an internet source backing me up:

www.whiteroachproductions.nl/FAQ.html


Posted by Subtle on Dec-27-2004 23:30:

ZxZDeViLZxZ: I have one important question in regards of your fabolous explanation, when a tune is "burned" onto a vinyl.. wouldnt that tune be burned right off the CD and the vinyl would get the same quality as the CD? (confuses me)


Posted by Dance123 on Dec-28-2004 05:03:

Exclamation

What I really would like know: what's the maximum frequency of mp3s (any difference in this between a 192 and 320kbps files) and vinyl. Anybody knows those numbers?!

Isn't it so that a CD sound brightest because it goes up to 22kHz and people can hear frequencies up to 20Hz (correct me if I am wrong) which a decent pair of hifi speakers can reproduce.

Anyway, I always get the impression that a 192kbps mp3 sounds less bright then a CDs, but I still don't know why that is!


Posted by Atlantis-AR on Dec-28-2004 14:04:

quote:
Originally posted by Dance123
What I really would like know: what's the maximum frequency of mp3s (any difference in this between a 192 and 320kbps files) and vinyl. Anybody knows those numbers?!

Isn't it so that a CD sound brightest because it goes up to 22kHz and people can hear frequencies up to 20Hz (correct me if I am wrong) which a decent pair of hifi speakers can reproduce.

Anyway, I always get the impression that a 192kbps mp3 sounds less bright then a CDs, but I still don't know why that is!


Yes, the maximum frequency depends on the bitrate, though it also varies slightly from file to file depending on the original frequency makeup. Use a spectrum analyser to find out the exact frequency and work out average values if you're really interested. From memory, a 128 kbps MP3 might throw away everything above 16 kHz, while a 192 kbps MP3 might have the LPF (lowpass filter) at 17 or 18 kHz. 320 kbps is probably around 20 kHz, or the LPF just has a gentler slope, can't remember.

Not only does compression to MP3 throw away the air frequencies, but it also screws up the rest of the frequency makeup, something you're definitely be able to hear after training your ears, and certainly if you've listened to CDs all your life and put on an MP3 track. And people can hear frequencies to about 16-18 kHz, yes. And depending on how youg or undamaged your ears are, this can go as high as 20-22 kHz,.

Well it sounds less bright because, one, the air frequencies above 17 kHz are discarded, and two, because the other frequencies (mostly audible in the highs) are affected too, adding 'swishy', 'flanging' artefacts that dull the sound and lose the sound's sparkle and transients.


Posted by ZxZDeViLZxZ on Dec-28-2004 15:43:

i would hope vinyl wasnt pressed from a cd source... i would hope itd be pressed from the highest quality format availble and not from a cd...


Posted by thecYrus on Dec-28-2004 19:05:

i made a short test..

original sample (ripped from cd)

converted sample (MP3 128kbit)

and now i reversed the mp3-file and played it in sync with the original, so that i'll only hear the difference.

and that's the result

you can see, that there are a lot of frequencies affected but like mentioned before the most frequencies are in the upper range.


Posted by Dance123 on Dec-29-2004 01:29:

Hi,

Great test, you can clearly hear the difference, but that's normal if you only use 128kbps. Perhaps you could do the same test with a 192kbps mp3 and perhaps also higher. Some say 192kbps sounds as good as CD, but I have my doubts, especially regarding high frequencies.

By the way, could you please explain the graphic, like what do the yellow and orange line mean? Also, your graphic seems to go only to 16khz so difficult to see what happens beyond that.

Last question: what's the name of that track and from which (compilation?) album did you take the sample. Sounds great, always looking for great new tunes!


Posted by thecYrus on Dec-29-2004 10:30:

quote:
Originally posted by Dance123
Last question: what's the name of that track and from which (compilation?) album did you take the sample. Sounds great, always looking for great new tunes!


Anjunabeats Volume Two
Smith & Pledger - Forever (Aspekt Remix)

i'll do the other test soon..


Posted by luizmenezesjr on Dec-29-2004 17:45:

Test with FFT

Hi Everyone,

Reading all posts, I make a test with WHITE NOISE (all frequencies with the same power)

The method was:

1-) Creating a source 24bits, 48kHz, Stereo WAV.
2-) Generating a WHITE NOISE with -1.0db (to avoid distortion)
3-) Saving the .wav as .mp3 in the formats:

Format - Size in bytes - Compression
Wav 24bits 48kHz - 288.046 - 0,00
MP3 128Kbps 44.1kHz - 18.288 - 15,75
MP3 128Kbps 48.0kHz - 18.432 - 15,63
MP3 192Kbps 44.1kHz - 26.856 - 10,73
MP3 192Kbps 48.0kHz - 27.072 - 10,64
MP3 320Kbps 44.1kHz - 43.392 - 6,64
MP3 320Kbps 48.0kHz - 43.993 - 6,55
MP3 VBR Highest 48kHz - 42.816 - 6,55

4-) Checking all the db frequencies on each file using a Fast Fourier Transform (FFT):
(click on the name format to download the mp3 or wav file)

Wav 24bits 48kHz


MP3 128Kbps 44.1kHz


MP3 128Kbps 48kHz


MP3 192Kbps 44.1kHz


MP3 192Kbps 48kHz


MP3 320Kbps 44.1kHz


MP3 320Kbps 48kHz


MP3 VBR Highest 48kHz


6-) Conclusion:
The MP3 Format removes high frequencies to compact the original wav source.
If you open any .mp3 format, you will notice a silence on the start and end of the file... So you may insert a silence on the start and end of the original wav to avoid loosing data.

If you want to distribute your files, you may use the MP3 320Kbps 48kHz format, because many high frequencies are still there after compacting.

But if you are working on 44.1kHz, the 192Kbps 44.1kHz is a good option too... Now it's up to you choose the best format...

You have to balance SIZE in bytes, and lost frequencies...

All my life I was using the Variable Bit Rate (VBR) Highest quality with 48kHz to compress my music, but now I see this format isn't good, because it looks like a MP3 128Kbps 48kHz but with low compression.

(sorry about my English)


Posted by Atlantis-AR on Dec-29-2004 18:37:

Great test, luizmenezesjr. But I've got a better option: use OGG Vorbis instead of MP3. It still discards the top-end, but when there are a lot of frequencies present here, it attempts to keep more of them.


Posted by State of Matter on Dec-29-2004 18:44:

quote:
Originally posted by ZxZDeViLZxZ
i would hope vinyl wasnt pressed from a cd source... i would hope itd be pressed from the highest quality format availble and not from a cd...


Think about how trance is produced man. From a whole bunch of digital audio. It is then exported to wav and the vinyl is pressed from this wav file, which is the equivalent of cd audio, 16 bit 44.1 khz. Sorry, but your argument doesn't hold water.


Posted by Orko on Feb-20-2005 19:52:

I am trying to convert some old tapes to CD.

I also wanted to record in 192khz/24bit, to maintain the highest possible quality.

One of my friends also wanted a copy, but wanted a normal cd, at 44.1khz/16bit,

Will the downsampling really make the recording sound bad? or is it just better to record in 44.1 in the first place?

Will recording in 44.1, and recording in 192 and then down sampling to 44.1, sound the same?


Posted by luizmenezesjr on Feb-21-2005 00:09:

Record your tapes using 48kHz/24bits...

It's a very good quality because the higher frequency that can be recorded is: 24.000 Hz (48kHz divide by 2), and humans can SOMETIMES hear 24kHz frequencies.

Is you record with 192kHz, you can keep your 96.000 Hz super high frequency, but we CAN'T hear it!, and I think your musics don't have any instrument that are using such high frequency.

(sorry about my English)


Posted by Derivative on Feb-21-2005 00:54:

further to the issues regarding vinyl. please take into account that the accuracy of the sound reproduced is variable and depends on many factors, not least of which include:

1) record wear (how many times the record has been played)
2) dust/lint on the surface of the vinyl
3) the frequency response of the stylus being used
4) stylus wear

the tracking weight of the stylus and the shape of the needle also influence record wear. a badly worn record sounds noticeably inferior. dust and lint on the surface of the record produce pops and clicks on playback and whilst some people like this characteristic of vinyl its not conductive to accurately representing the sound of the master recording. additionally, lint on the surface of the vinyl can effect tracking and if a stylus doesnt track properly in the groove you can introduce quite severe record wear since the needle exerts massive pressure on the groove if not properly aligned (this again also depends on the tonearm configuration and the shape of the stylus).

record wear is also downward exponential in a sense. the first few plays are were the greatest amount of wear occurs. after a hundred or so plays subsequent degradation is not noticeable although by that point you will probably have new records and will be rinsing those instead.

all told cd is generally a more reliable medium for retaining quality of the original disk recording because it does not experience wear through play and the frequency response of a laser lens is generally cannot generally be compromised in the same way in which a stylus wears over time.


Posted by Derivative on Feb-21-2005 01:02:

quote:
It's a very good quality because the higher frequency that can be recorded is: 24.000 Hz (48kHz divide by 2), and humans can SOMETIMES hear 24kHz frequencies.

Is you record with 192kHz, you can keep your 96.000 Hz super high frequency, but we CAN'T hear it!, and I think your musics don't have any instrument that are using such high frequency.


24,000 hz is largely inaudible. you would have to play the signal at fairly high amplitude before you become aware that it is there and you typically notice high frequency sound at high amplitude because it devastates your ears. for all intents and purposes there is very little noticeable difference between a 48,000hz recording and a 44,100hz recording. dj thy mentioned something about 48,000hz recordings before and it would be cool if he could pop in and reiterate some of it since i forgot what it was. but it was informative.

with a 96,000hz recording and nyquest theorum the highest reproducible frequency is 48,000hz which is far and away above the limits of human hearing. and in either case those frequencies about 20,000 hz typically arent driven at very high amplitudes so it becomes very difficult to notice that they are there. i sometimes swear i only notice the difference because i have a bunch of spectrum analysers open and am actively listening for differences in each recording. in club? nobody will give a damn. base graffiti's house always wins will devastate slinky even if its played on a cassette recorder.


Posted by luizmenezesjr on Feb-21-2005 04:12:

MAYBE you can hear 24.000 Hz, I can't, for me is just a little noise with high pitched with very low gain in comparison with lower frequencies... (I lost some ear capabilities of my right ear listening to high music when I Dj sometimes a go ..

But anyway... do you have an speaker capable of reproduce 24.000Hz??

By example...:

"Your equipament plays only until 22.000 Hz". If you plays a 24.000 Hz sound, it will be aliasing to 20.000 Hz (think it like a mirror, when the frequency reachs the maximum Hz, it starts to play inversed - or subtracting).

So if your equipament reachs 22kHz, the 20kHz and 24kHz will be the same sound... make a test... This happens because mechanical characteristics of the speakers and tweeters

Many Many speakers and tweeters CAN'T play higher frequencies than 22kHz (specials speakers can)....

Check the Mackies Studio Monitors (i.e. HR824), one of the best on the market, above 20kHz the gain starts to drop (see specs on their website).

With Frequency Meters, like the ones you are using, you can detect such frequencies, but hear it you have to be careful with your speakers and maximum acepted frequencies.

(just my oppinion

(sorry about my English)


Pages (2): [1] 2 »

Powered by: vBulletin
Copyright © 2000-2021, Jelsoft Enterprises Ltd.